The world's most clever kitty cat
Merge pull request #6 from loco-choco/main
Added learning from pdf (admin only)
authored by
GitHub
a985794c
91aadb17
+2
-1
pyproject.toml
+2
-1
pyproject.toml
+23
src/bingus/cogs/markov.py
+23
src/bingus/cogs/markov.py
···
4
4
import discord
5
5
import pytesseract
6
6
import PIL
7
+
import pypdf
7
8
from discord.ext import commands
8
9
from discord.message import Message
9
10
from pathlib import Path
···
95
96
await ctx.respond(head, file=discord.File(fd, filename="weights.txt"))
96
97
else:
97
98
await ctx.respond(f"{head}:\n{msg}")
99
+
100
+
@require_owner
101
+
@commands.slash_command()
102
+
async def pdf(
103
+
self, ctx: discord.ApplicationContext, file: discord.Option(discord.Attachment)
104
+
):
105
+
await ctx.defer(ephemeral=True)
106
+
raw = await file.read()
107
+
try:
108
+
pdf = pypdf.PdfReader(io.BytesIO(raw))
109
+
# i = 0
110
+
for page in pdf.pages:
111
+
# i += 1
112
+
# printf("Bingus learned a page! {i}/{pdf.get_num_pages()}")
113
+
text = page.extract_text()
114
+
self.markov.learn(text)
115
+
await self.update_words()
116
+
await ctx.respond(
117
+
"> Bingus learned something from the pdf!", ephemeral=True
118
+
)
119
+
except pypdf.errors.PdfReadError:
120
+
await ctx.respond("> Bingus only understands pdf files!", ephemeral=True)
98
121
99
122
@require_owner
100
123
@commands.slash_command()
+11
uv.lock
+11
uv.lock
···
90
90
{ name = "msgpack" },
91
91
{ name = "pillow" },
92
92
{ name = "py-cord" },
93
+
{ name = "pypdf" },
93
94
{ name = "pytesseract" },
94
95
{ name = "python-dotenv" },
95
96
]
···
100
101
{ name = "msgpack", specifier = ">=1.1.0" },
101
102
{ name = "pillow", specifier = ">=11.1.0" },
102
103
{ name = "py-cord", specifier = ">=2.6.1" },
104
+
{ name = "pypdf", specifier = ">=5.3.1" },
103
105
{ name = "pytesseract", specifier = ">=0.3.13" },
104
106
{ name = "python-dotenv", specifier = ">=1.0.1" },
105
107
]
···
360
362
sdist = { url = "https://files.pythonhosted.org/packages/27/c7/c539d69d5cfa1ea5891d596212f73d619e40c7fc9f02ae906f4147993b94/py_cord-2.6.1.tar.gz", hash = "sha256:36064f225f2c7bbddfe542d5ed581f2a5744f618e039093cf7cd2659a58bc79b", size = 965087 }
361
363
wheels = [
362
364
{ url = "https://files.pythonhosted.org/packages/e7/90/2690ded84e34b15ca2619932a358c1b7dc6d28fe845dfbd01929fc33c9da/py_cord-2.6.1-py3-none-any.whl", hash = "sha256:e3d3b528c5e37b0e0825f5b884cbb9267860976c1e4878e28b55da8fd3af834b", size = 1089154 },
365
+
]
366
+
367
+
[[package]]
368
+
name = "pypdf"
369
+
version = "5.3.1"
370
+
source = { registry = "https://pypi.org/simple" }
371
+
sdist = { url = "https://files.pythonhosted.org/packages/da/5b/67df68ec4b934aae9ca89edfb43a869c5edb3bd504dd275be9e83001d3e9/pypdf-5.3.1.tar.gz", hash = "sha256:0b9b715252b3c60bacc052e6a780e8b742cee9b9a2135f6007bb018e22a5adad", size = 5011845 }
372
+
wheels = [
373
+
{ url = "https://files.pythonhosted.org/packages/f4/0c/75da081f5948e07f373a92087e4808739a3248d308f01c78c9bd4a51defa/pypdf-5.3.1-py3-none-any.whl", hash = "sha256:20ea5b8686faad1b695fda054462b667d5e5f51e25fbbc092f12c5e0bb20d738", size = 302042 },
363
374
]
364
375
365
376
[[package]]